Does inward remittance influence the choice of households’ cooking fuel?

ABSTRACT

Among the Sustainable Development Goals is to ensure the consumption of clean and affordable energy by all by the close of 2030. Over the last decade, calls for energy transition from unclean energy to clean ones have intensified. Consequently, the price of clean energy has been subsidised in Ghana but this is not enough as energy transition still entails considerable costs for households. The receipt of inward remittance improves households’ income which can aid their transition but literature in this regard is scanty. Applying the Special Regression Method to household data from the Ghana Living Standards Survey, we examine how the receipt of income remittance influences the choice of households’ cooking fuels. We found an increase in income remittance increases the odds of households adopting clean cooking fuels. We recommend that the government of Ghana undertake policies that improve household income to encourage the adoption of clean fuels and reduce environmental pollution.

KEYWORDS:

• Cooking fuels
• energy transition
• Ghana
• households
• logit model
• remittances

JEL CLASSIFICATION:

• C25
• Q23
• Q42

1. Introduction

The seventh goal of the Sustainable Development Goals (SDGs) is to ensure the consumption of clean and affordable energy by 2030. The reason is that the consumption of unclean energy from fossil fuel sources such as firewood, charcoal, and kerosene lead to environmental problems such as air pollution, deforestation, and soil erosion. The burning from the use of solid-based fuels releases harmful gases such as CO₂, methane, and nitrogen oxides into the atmosphere. When the assimilative capacity of the environment to absorb these harmful gases is exceeded, they remain within the atmosphere leading to climate change. Moreover, the consumption of unclean energy poses health risks that can be life-threatening. That is, the release of harmful gases from the use of unclean energy threatens the capacity of the environment to sustain the lives of humans and animals.

Younger et al. (2022) showed in a systematic review that exposure to harmful gases from using dirty fuels reduces the birth weight of newborn babies, and leads to stillbirths and neonatal mortality in low-income countries. In Ghana, Amegah et al. (2012) found the same evidence in the national capital using information on 592 mothers and their newborn babies. This contributes to global mortality and morbidity as dirty fuel-induced deaths are premature and can be as high as 3.2 million deaths per year (World Health Organization [WHO] 2022). Also, the continuous use of unclean energy is expected to contribute about 40 billion tons of CO₂ emissions by the close of 2050 (Gani 2021). Several shreds of evidence exist on the damaging effect of polluted air on the health of Ghanaians resulting from the use of fossil fuels (Zhou et al. 2011; Jack et al. 2015 & Weber et al. 2020). According to WHO (2022), the majority of Ghanaians were exposed to air pollution as emissions such as harmful particulate matter (P2.5) exceeded their accepted levels.

The consumption of unclean energy in Ghana is motivated by the high level of energy poverty according to Adusah-Poku and Takeuchi (2019). Energy poverty according to this study refers to the lack of or inadequate access to modern, clean, and affordable energy by households compelling them to use fuels from dirty and inefficient sources for their domestic and commercial activities. Energy poverty motivates the use of dirty fuels because the energy poor without access to modern and clean fuels resort to dirty fuels for their energy needs. The lack of access is explained by factors such as low levels of income, high prices of modern energy, and the location of the household (Twumasi et al. 2021). The level of energy poverty in Ghana stood at 82.5% (Adusah-Poku and Takeuchi 2019). This means 4 out of 5 households in Ghana rely on inefficient sources of energy for their daily activities. This revelation confirms statistics from the Ghana Statistical Service [GSS] in 2017 on the prevalence of energy poverty. Table 1 shows about 80 percent of Ghanaian households rely on unclean sources of energy such as cut wood and charcoal while less than 20 percent use clean fuels such as electricity and gas for cooking. The danger of this high energy poverty is that it might lead to the depletion of fossil fuels in addition to the associated health and environmental risks (Alem et al. 2016; Meng et al. 2021; Twumasi et al. 2021). This makes it difficult to fight poverty as poverty and environmental degradation tend to feed each other (Ogbeide-Osaretin 2021).

Table 1. Main type of fuel used by Ghanaian households for cooking.

Fuel Type	Frequency	Percent	Cumulative percent
Wood	6,511	49.20	49.20
Charcoal	3,945	29.81	79.00
Gas	2,573	19.44	98.44
Electricity	30	0.23	98.67
Kerosene	9	0.07	98.74
Crop residue	134	1.01	99.75
Sawdust	2	0.02	99.77
Other	10	0.08	100.00

Source: Ghana Living Standard Survey [GLSS] (2017).

Consequently, policymakers have strengthened calls for households to abandon the use of dirty fuels for clean ones such as Liquified Petroleum Gas (LPG), electricity, and solar for their commercial and domestic needs (Olanrewaju et al. 2019; Bekun et al. 2021). To encourage uptake, the government of Ghana initiated policies such as subsidising the price of LPG but this is not enough as energy transition still entails considerable costs for households (Karimu, Mensah, and Adu 2016). It requires considerable financial commitment for households to adopt and continue to use clean energy. For instance, the use of LPG requires initial money to acquire the gas cylinder and stove as well as the regular expenditure on gas refills. The use of solar and electric stoves also requires similar financial commitments. Compared to fossil fuel options such as firewood and charcoal, the majority of households with limited income will not want to transition. Based on this, any factor that improves household income can aid them to shift to cleaner fuels and reduce environmental pollution. Besides, some households will choose to use a combination of clean and unclean fuels due to energy security, fuel availability, and preference as explained by the fuel stacking hypothesis (Taylor et al. 2011; Karakara, Osabuohien, and Asongu 2021).

A key factor that is gaining attention in the energy transition debate is remittance. Remittance refers to the inflow of financial and material payments from individuals who have permanently or temporarily migrated to other countries (Alleluyanatha and Treasure 2021). The receipt of income remittance by Ghanaian households as of 2017 averaged GHS 369.74 with a standard deviation of GHS 1977.13 showing a lot of variation. As of 2022, personal remittances received in Ghana were $4.7 billion. Further statistics indicated that income remittance received in Ghana peaked at $5 billion in 2015 (Statista, 2023a). Also, according to the Developing Markets Associates (2018), Ghana was ranked as the 2nd largest recipient of inward remittance in sub-Saharan Africa (SSA). In the same year, Ghana received in total of $3.8 billion in remittances representing 7.4% of the country’s income. The forecast by experts shows that receipts of remittances will continue to increase. The understanding is that income remittance represents an extra source of funds for households which are used to improve their standard of living. This stems from the fact that funds from inward remittances are used for consumption and investment purposes such as opening new lines of businesses or boosting existing ones (Ajefu 2018; Alleluyanatha and Treasure 2021). Other reasons for remittance include loan repayment, education, and bequest purposes (Lucas and Stark 1985; Rapoport and Docquier 2006; Ajefu 2018).

Several schools of thought exist on the role of remittance in energy transition. The supporters of the remittance-led pollution hypothesis argued that inward remittances increase household income, which is mostly consumed leading to an increase in aggregate consumption. Firms respond by increasing industrialisation which increases environmental pollution (Rahman, Cai, and Ahmad 2019). The other channel via which remittance contributes to pollution is that the inflow of remittance increases both the extensive and intensive margins of savings (Osei-Gyebi et al. 2023). This increases credit creation by financial institutions which in turn increases industrialisation and pollution at the same time (Rahman, Cai, and Ahmad 2019; Ahmad et al. 2019). The consensus from these studies indicates that inward remittances promote environmental pollution via increases in aggregate expenditure and credit creation by financial institutions.

Table 2. Summary description of variables.

Variable	Symbol	Description	Mean	SD	Min.	Max.
Fuel choice	FC	The choice of fuel used by the household	0.14	0.349	0	1
Income remittance	IR	The amount of income remittance received	369.7	1,977	0	100000
Household wage	HW	The wage received by households	3,574	9,512	0	216000
Household size	HS	The size of the household	6.14	3.54	1	28
Location (rural)	LOC	The location of the household	0.636	0.48	0	1
Poverty (poor)	PS	The poverty status of the household	0.198	0.39	0	1

Notes: Author’s construction based on data from GLSS (2017), SD is standard deviation, min., and max. represents minimum and maximum, respectively; Income remittance and household wage measured in Ghana Cedis (GHS).

Opposing views suggest that inward remittance serves to fill the financial gap from inadequate domestic savings just like foreign direct investments and external aid. The additional funds are allocated to the adoption of efficient technologies and renewable energy consumption that reduces environmental pollution (Islam 2022). This view suggests that inward remittance facilitates the energy transition process which reduces environmental pollution. However, there are two drawbacks to this view. First, it is expressed at the macro level and who receives what is difficult to identify. The second criticism is that the pathway from inward remittance to pollution abatement is absent or not clear. This is because most policies or strategies targeted at energy transition and pollution abatement in developing countries like Ghana are sponsored with external funds from development partners like the World Bank and the United Nations (UN). For example, the renewable energy plan in Ghana was funded with funds from China.

The current study indicates that remittance can contribute to a reduction in environmental pollution when the remittance-pollution nexus is examined at the micro level. We argue that an increase in household income due to inward remittance can influence the choice of fuel used by households for their activities. First, the inflow of remittance increases household income which allows them to afford cleaner energy like LPG. Also, the investments in businesses due to remittance increases the time and opportunity cost of using unclean energy such as firewood. For instance, the time households spend collecting firewood for cooking can be used for their businesses by using a gas stove. Lastly, inward remittance raises the social status of households which may encourage them to adopt cleaner fuels. That is, the receipt of remittance can improve the economic and social conditions of households facilitating the adoption of cleaner fuels and reducing environmental pollution.

Yet, no study has examined the potential of inward remittance to reduce environmental pollution by influencing the fuel choice of households. Also, the literature is replete with several studies such as Mensah and Adu (2015), Karimu, Mensah, and Adu (2016), Alem et al. (2016), and Bofah, Appiah-Konadu, and Ngwu (2022) that examined the drivers of energy choice of Ghanaian households in different contexts but none of them considered the important role of inward remittance. Based on the damaging effects of using dirty cooking fuels on the health and environment of households as well as the growing importance of inward remittance for households, a study that examines the role of inward remittance on the energy choice of households will be essential for the energy transition drive in Ghana. We contribute to the literature by examining the effect of inward remittance on the fuel choice of households in Ghana. Consequently, the objectives of the study are twofold. The first objective is to estimate the effect of income remittance on the fuel choice of households in Ghana. Secondly, we estimate the effects of other socioeconomic factors on the fuel choice of households in Ghana.

In light of the findings, there is a strong connection between the living standards of households and their choice of cooking fuels and by extension their ability to transition and reduce environmental degradation. Households that have received income remittance, have a wage income, are non-poor, and reside in urban areas are more likely to adopt clean cooking fuels. On the contrary, households that are large, poor, and reside in rural areas are less likely to choose clean fuels. It can therefore be said that an improvement in the living standards of Ghanaian households is important for them to use clean fuels and reduce environmental degradation. Government policies on energy transition must be focused on improving the livelihoods of poor and rural households as that will go a long way to increase the use of clean fuels and reduce environmental degradation.

The rest of the study is organised as follows. Section two reviews the relevant literature on the subject; Section three presents materials and methods adopted by the study; Section four presents discussions of the empirical results, and Section five presents conclusions, policy implications, and limitations of the study.

2. Literature review

2.1 Theoretical review

The remittance-led pollution hypothesis explains that the inflow of income remittance leads to higher levels of environmental pollution. This is explained via the aggregate consumption and credit creation channels (Rahman, Cai, and Ahmad 2019; Ahmad et al. 2019). The inflow of remittance improves household income directly and indirectly through the income-generating businesses that it allows. According to the fundamental psychological law by Keynesians, households’ consumption is proportional to their income. It follows that a considerable fraction of the income increase is consumed which increases total consumption and aggregate expenditure. Firms in a bid to make more profit respond by expanding their production which increases industrialisation. On the other hand, the credit creation channel suggests that the receipt of income remittance increases the savings of households. This is seen in both the extensive margin relating to the propensity to save (Osei-Gyebi et al. 2023) and the intensive margin relating to the amount saved by households (Wolff 2015). The increase in bank deposits from higher savings drives interest rates down which increases credit creation for businesses to invest. The result is an increase in production and industrialisation which is accompanied by environmental degradation. While the increasing effect of income remittance on environmental pollution has received a lot of attention, its reducing effect, especially at the micro level remains largely unexplored. We argue that the receipt of income remittance by households can reduce environmental pollution by examining how it influences the choice of household cooking fuels.

The study is also explained by the energy ladder hypothesis which describes the relationship between household income and their choice of cooking fuels. The hypothesis suggests that households with low incomes cannot afford clean and modern fuels so they rely on dirty fuels which are mostly fossil-fuel based. On the contrary, households with high incomes use clean fuels mainly because they have the means to afford them. This means an improvement in household income empowers them to move up the energy ladder shifting away from dirty fuels to clean and modern ones (Van der Kroon, Brouwer, and Van Beukering 2013; Karakara, Osabuohien, and Asongu 2021). Hence, the energy ladder hypothesis postulates a positive relationship between income level and the use of clean fuels by households. However, the energy ladder hypothesis has been criticised on the basis that the energy choice of households is not determined by only income because several social, demographic, and institutional factors play key roles (Masera, Saatkamp, and Kammen 2000). Another criticism is that households do not make a complete shift/switch from one type of fuel to another giving birth to the fuel stacking and fuel switching hypotheses (Taylor et al. 2011; Karakara, Osabuohien, and Asongu 2021).

While fuel stacking explains that households use different types of fuel at the same time, fuel switching suggests that households resort to a particular type of fuel at a point in time depending on the kind of food being prepared, preferences, fuel price, and availability. This implies households switch between different types of fuels based on prevailing conditions at a particular point in time (Alem et al. 2016). About the current study, the receipt of inward remittance improves household income and has the potential to alter their choice of fuel used for cooking besides the already established factors by the extant literature. The improved incomes may encourage the shift to cleaner fuels without displacing traditional cooking fuels (Meng et al. 2021). The key point is that the complete or partial shift necessitated by remittance reduces the intensity of using fossil-based fuels which is essential as far as environmental sustainability is concerned. Given the increasing importance of remittance for households in developing countries like Ghana, we deem it relevant to fill the knowledge gap by investigating how the receipt of income remittance influences the choice of cooking fuels by households.

2.2 Empirical review

The extant literature has examined the determinants of households’ choice of cooking fuels in different contexts and jurisdictions. For instance, Heltberg (1999) using household data from Guatemala found the time and opportunity costs of using firewood as a critical factor apart from income in determining the choice of household cooking fuels. Similar to findings in the extant literature, the study observed that modern cooking fuels do not completely displace traditional fuels suggesting the existence of fuel stacking and switching behaviour among households in Guatemala. Also, based on microdata on rural households in China, Hou et al. (2017) examined the issues of energy poverty and energy transition. Per their findings, energy poverty tends to dominate in rural areas where more than half of households use dirty fuels. However, energy poverty tends to be relatively low (5%) in urban areas. Further evidence shows socioeconomic factors also influence the energy choice of households in rural China. Paudel, Khatri, and Pant (2018) also found similar determinants of household energy choice in Afghanistan making use of a multinomial logit model. Additional factors such as wealth and electricity access were noted to influence the choice of households in Afghanistan to adopt LPG.

Mekonnen and Köhlin (2009) and Alem et al. (2016) investigated the determinants of the choice of cooking fuels among urban households in Ethiopia. Their findings revealed fuel price, education, and economic status as key factors that inform the choice of household cooking fuels. The study also found support for the fuel stacking behaviour in Ethiopia. Specifically, Mekonnen and Köhlin (2009) found that households not only increase the use of modern fuels in response to income increase but also the use of traditional fuels such as wood and charcoal. In all, their results show that higher incomes allow households to increase the total number of fuels used at home. Using the rural areas in the Enugu State of Nigeria, Nnaji, Ukwueze, and Chukwu (2012) found socioeconomic features of households such as their income, age, education, and cooking setup as the factors that influence the choice of their cooking fuel. The study stressed the predominance of fuelwood as the major fuel type among rural households in the study area.

In Ghana, Amegah et al. (2012) using data on patients from Accra found that the use of charcoal exposes pregnant women to air pollution which reduces the birth weight of newborn children. The magnitude of this reduction increases if mothers are exposed to both indoor pollution and pollution from the burning of garbage. Applying multinomial logit to the GLSS data, Mensah and Adu (2015) identified social, economic, and demographic factors as the key factors that inform the decision of households to shift from fuelwood to LPG in Ghana. The study stressed the availability and price of LPG as vital for households to make the shift. Narrowing the focus to just one type of fuel, Karimu, Mensah, and Adu (2016) investigated the factors that influence the decision of Ghanaian households to adopt LPG. Using the GLSS data in a parametric analysis, the study revealed households’ access to infrastructure as a key determinant for choosing LPG in addition to the known socio-economic factors such as education, income, and household size. While there is no difference in the decision to use LPG over time, the location of households is an important driver of their decision to adopt LPG.

Also, Meng et al. (2021) used data from Accra, Takoradi, and Tamale to examine the use of different fuels by rural households. Just like previous studies, income, social, and demographic factors were found to influence the intensity of fuel use by households. Particularly, Meng et al. (2021) indicated that increases in household size increase the propensity for households to use fossil-based fuels like charcoal and firewood. Their findings also revealed significant effects in the choice of cooking fuels based on households’ location as residents within Tamale tend to use more fuelwood and charcoal compared to those in Accra and Takoradi. Furthermore, Martey et al. (2021) using the GLSS datasets focused the discussion on the choice of cooking fuel among the time and consumption poor in Ghana. According to their findings, compared to the non-poor, the consumption poor are more likely to adopt solid fuels which are unclean and inefficient. The study suggests relevant institutions in Ghana strongly pursue the Sustainable Energy for All (SE4ALL) agenda to encourage the adoption of clean fuels and reduce poverty.

Shifting the analysis to the health consequences, Twumasi et al. (2021) investigated the drivers of household cooking fuels in Ghana. Apart from confirming the socio-economic factors as determinants of household energy, the findings identified credit, internet access, assets, and a relative living in an urban area as key drivers of energy choice for Ghanaian households. Additional findings from their inquiry indicate that the use of clean cooking fuel can improve the health of households with the effect on females being more pronounced than their male counterparts. Again, Dongzagla and Adams (2022) using data from the 2014 Demographic Health Survey identified household wealth as the major driver among the social, demographic, and economic factors that determine the choice of cooking fuel among urban households in Ghana.

Likewise, Bofah, Appiah-Konadu, and Ngwu (2022) identified additional factors such as the dwelling of household, employment, and dependency burden as important factors that explain the transition to clean fuels in Ghana. Precisely, households with paid employment, modern house, and low dependency burden are more likely to adopt clean fuels. Their study recommended that energy transition policies in Ghana should target improving housing and employment opportunities for households to achieve maximum results. Most recently, Adjei-Mantey and Takeuchi (2023) considered the role of risk aversion in determining energy choices among Ghanaian households. Using the GLSS data, their findings indicate that households which are risk-averse are less likely to choose LPG for their cooking. The presence of appropriate infrastructure and public education can help mitigate the risk and encourage households to use it.

Another group of empirical studies investigated the effect of remittances on different indicators. For example, Karmaker et al. (2023) investigated the effect of remittance on the consumption of renewable energy using data on the top destination countries of remittances from 1990 to 2018. Their findings show that receipt of remittance has a positive effect on the consumption of renewable energy by households in these countries. Government policies must be directed at incentivizing households to pay for their renewable energy with the remittance received. Hosan et al. (2023) also investigated the effect of remittance on multidimensional energy poverty making use of a national household survey. Findings revealed that households receiving remittance tend to have a lower level of multidimensional energy poverty. This evidence adds to a growing body of literature that emphasises the importance of remittance to the livelihoods of households. The study recommended policies that promote and support migrant workers to improve the volume of remittances received. Rahman et al. (2021) also analyzed the effect of remittance on energy consumption for selected countries in South Asia. Concentrating on the top four remittance-receiving countries in South Asia from 1976 to 2019, the findings of the study showed a long-run association between energy consumption and the receipt of income remittance. Specifically, remittance has a positive effect on energy consumption for the period under study. This confirms the assertion that policies to enhance the inflow of remittance will increase energy consumption.

Based on the review of the empirical literature, several factors have been identified as drivers of energy choice for households both in Ghana and elsewhere. Social, demographic, and economic factors such as income, education, marital status, location, fuel price, and energy availability were identified. While some studies concentrated on just one type of fuel like LPG, others investigated the drivers for different fuel types. The positive effect of remittance on households and the economy has also been documented. The goal is to realise the SE4ALL by encouraging the adoption of clean fuels to reduce the dependence on solid-based fuels. One single theme that runs through this review is that improvement in household income and living standards is important for this transition. The receipt of income remittance improves the economic and social conditions of households but its potential to help achieve the SE4ALL target remains unexplored. We bridge this gap by examining how the receipt of income remittance influences the choice of cooking fuels for households in Ghana.

3. Materials and methods

3.1 Source of data and variable description

The study relies on microdata on households in Ghana from the most recent round of the Ghana Living Standards Survey (GLSS) ) conducted in 2017 by the Ghana Statistical Service (GSS). The GLSS is the most comprehensive survey on households in Ghana consisting of important information about their living standards and characteristics. Specifically, the study used information such as the main fuel used by households for cooking, the amount of income remittance received, the household size, the amount of household wage, their location, and poverty status for its analysis. After cleaning and merging appropriate data sheets, information on some 4,769 households across the various regions of Ghana was used for the analysis. The summary statistics for variables used for the study have been presented in Table 2.

The dependent variable of the study is a response to the question of the main fuel used by households for cooking. The fuel types include firewood, charcoal, LPG, kerosene, electricity, sawdust, and crop and animal residue. To simplify the analysis, the different types of fuels were grouped into dirty and clean fuels. The dirty fuels include fuels from fossil sources such as firewood, charcoal, kerosene, sawdust, and crop and animal residue. On the other hand, clean fuels include LPG and electricity. The focus of the study was to examine how inward remittance encourages households to shift from unclean fuels to cleaner ones. Hence, the dirty fuel group was coded as 0 and the clean fuel as 1. This allows an analysis of how the receipt of inward remittance influences the likelihood of households to adopt clean fuels. The emphasis of this study is on the extensive margin of using clean fuels rather than the intensive margins as the focus is on energy transition.

The main independent variable is the amount of income remittance received by households. The amount of income remittance ranged from 0 to GHS100,000. Empirical findings such as Mensah and Adu (2015), Karimu, Mensah, and Adu (2016), and Bofah, Appiah-Konadu, and Ngwu (2022) noted household income as the main determinant of household choice of energy so any factor such as the receipt of remittance which improves household income encourage the use of clean fuels. The expectation is that an increase in the receipt of income remittance increases the propensity for households to use clean fuels.

The study controls for the wage of the household, their size, location, and poverty status. The main barrier to the adoption of clean fuels by households is that they are relatively expensive. It follows that households with high wages will be more likely to use clean fuels because they can afford them. Also, the energy ladder hypothesis explains that households shift to cleaner fuels as their wages increase. This means households with low wages tend to use dirty fuels while those with high wages use clean fuels (Alem et al. 2016; Bofah, Appiah-Konadu, and Ngwu 2022). Hence, an increase in household wage is expected to increase their likelihood to use clean fuels for cooking.

Household size measures the total number of individuals for which the household head is responsible. This mainly includes the children and spouse of the head plus other relatives that may be under their care. The high dependency burden associated with larger households implies expenditure per member of the household is low holding constant their level of income. This generally reduces the purchasing power of such households and makes them less likely to choose clean fuels (Meng et al. 2021). This suggests a positive relationship between household size and the likelihood to choose dirty fuels, holding all other factors constant. It is therefore expected that larger households are more likely to use dirty fuels compared to smaller households.

Location measures the geographical area of the household. Basically, it tells whether the household resides in a rural or urban area. GLSS denotes households in rural areas as 1 and 2 for those in urban areas. Without discounting urban poverty, poverty tends to be endemic and systemic in rural areas. Generally, rural-urban drift is mostly driven by a lack of economic opportunities in rural areas and the hope that such opportunities can be found in urban areas. Besides, the lower opportunity cost of rural dwellers due to high unemployment and underemployment means they have enough time to collect fuelwood which is readily available in these areas (Nnaji, Ukwueze, and Chukwu 2012). It is therefore expected that households in rural areas with lower income will be less likely to use clean fuels.

Poverty is a multifaceted phenomenon that includes severe deprivation in the material, social, physical, and psychological aspects of the individual. The implication is that many factors contribute to making one poor and these factors are mostly interlocked (World Bank 2022). The Ghana Living Standard Survey (GLSS) categorises households as either non-poor, poor, or very poor based on the consumption expenditure per adult equivalent (GSS, 2017). To correctly control for the poverty status of the household, this variable was recoded as 1 for the poor and very poor households and 0 for non-poor households. This was consistent with studies such as (Nnaji, Ukwueze, and Chukwu 2012). The study expects non-poor households to be more likely to choose clean fuels such as LPG and electricity whereas poor households are more probable to choose dirty fuels. This is due to the fact that, unlike poor households, non-poor households can afford the relatively higher cost associated with using clean fuels.

3.2 Empirical model

The purpose of the current research is to examine how the receipt of income remittance influences the choice of cooking fuel by households in Ghana. The theoretical bedrock of this study is the energy ladder hypothesis which postulates that households with low income tend to consume unclean fuels. However, they move up the fuel ladder to consume clean fuels when they experience increases in their income (Perman et al. 2003). Although several other factors such as preference, fuel price, and fuel availability inform the choice of cooking fuel, the main determinant of cooking fuel is household income (Alem et al. 2016; Karimu, Mensah, and Adu 2016; Bofah, Appiah-Konadu, and Ngwu 2022). The current study recognises that the receipt of income remittance improves household income and might quicken their ascension up the fuel ladder where they consume cleaner fuels. Consequently, we regress the fuel choice of households on their receipt of income remittance and control for household characteristics such as their wage, size, location, and poverty status. This is specified as; (1) $F{C}_i={\text{β}}_0+{\text{β}}_1\mathrm{l}\mathrm{n}\left(I{R}_i\right)+{\text{β}}_2\mathrm{l}\mathrm{n}\left(H{W}_i\right)+{\text{β}}_{3}H{S}_i+{\text{β}}_{4}LO{C}_i+{\text{β}}_{5}P{S}_i+{\mu }_t$(1) where $F{C}_i$ is defined as the choice of fuel by household i which is a binary variable coded as 0 (dirty fuel) and 1 (clean fuel), $I{R}_i$ is defined as household income from remittance, $H{W}_i$ as the wage of the household, $H{S}_i$ as household size, $LO{C}_i$ as the location of the household, and $P{S}_i$ as the poverty status of the household. ${\text{β}}_0$ is the intercept, ${\mu }_t$ as the stochastic error term and ${\text{β}}_1$ to ${\text{β}}_5$ as coefficients of the independent variables.

3.3 Estimation strategy

According to Allison (1999), a dichotomous dependent variable violates the homoscedasticity and normality requirements of the linear regression model. As a result, neither the coefficient estimates nor the standard error estimations accurately represent genuine standard errors. Moreover, using the ordinary least squares method to estimate a linear probability model will result in predicted values that fall outside the probability range of probable values (0,1). Based on the fact that the dependent variable is coded 0 for dirty fuel and 1 for clean fuel, the logistic regression model was considered appropriate for this study. This model converts probability to odds before taking the logarithm of odds. By doing this, the lower and upper bounds of the probability are eliminated (Greene 2008). The logistic regression model is as follows: (2) $logit\left(F{C}_i\right)=\log \left\{{\displaystyle \frac{F{C}_i}{1-F{C}_i}}\right\}={\text{β}}_0+{\text{β}}_{1}I{R}_i+{\text{β}}_2\sum _{i=1}^n{X}_i$(2) ${\text{β}}_0$ and ${\text{β}}_1$ are the regression parameters, $I{R}_i$ is the main independent variable, $X_i$ represents the control variables including their wage, size, location, and poverty status. $F{C}_i$ denotes the proportion of households using clean fuels in the past year. For the linear model, the marginal effect of the variables estimated at certain values of the explanatory variables is of interest. Therefore, Equation 1(1) $F{C}_i={\text{β}}_0+{\text{β}}_1\mathrm{l}\mathrm{n}\left(I{R}_i\right)+{\text{β}}_2\mathrm{l}\mathrm{n}\left(H{W}_i\right)+{\text{β}}_{3}H{S}_i+{\text{β}}_{4}LO{C}_i+{\text{β}}_{5}P{S}_i+{\mu }_t$(1) can be rewritten as; (3) $logit\left(F{C}_i\right)={\text{β}}_0+{\text{β}}_1\mathrm{l}\mathrm{n}\left(I{R}_i\right)+{\text{β}}_2\mathrm{l}\mathrm{n}\left(H{W}_i\right)+{\text{β}}_{3}H{S}_i+{\text{β}}_{4}LO{C}_i+{\text{β}}_{5}P{S}_i+{\mu }_t$(3) Equation 3(3) $logit\left(F{C}_i\right)={\text{β}}_0+{\text{β}}_1\mathrm{l}\mathrm{n}\left(I{R}_i\right)+{\text{β}}_2\mathrm{l}\mathrm{n}\left(H{W}_i\right)+{\text{β}}_{3}H{S}_i+{\text{β}}_{4}LO{C}_i+{\text{β}}_{5}P{S}_i+{\mu }_t$(3) is the estimated model to determine the influence of income remittance on the choice of cooking fuel used by Ghanaian households. All the variables in Equation 3(3) $logit\left(F{C}_i\right)={\text{β}}_0+{\text{β}}_1\mathrm{l}\mathrm{n}\left(I{R}_i\right)+{\text{β}}_2\mathrm{l}\mathrm{n}\left(H{W}_i\right)+{\text{β}}_{3}H{S}_i+{\text{β}}_{4}LO{C}_i+{\text{β}}_{5}P{S}_i+{\mu }_t$(3) are as explained before. As previously explained, we expect an increase in the receipt of income remittance to increase the likelihood of households using clean cooking fuels like electricity and LPG. The inflow of remittance improves household income increasing their propensity to choose clean fuels.

The study recognises the possibility of the main explanatory variable being endogenous because several factors explain the decision for people to migrate and remit. This means the receipt of income remittance is not exogenous and requires appropriate methods to estimate valid effects. Since the study considers a binary choice model with an endogenous explanatory variable (EEV), the Special Regression Method (SRM) by Dong and Lewbel (2015) was used in this regard. The SRM assumes that the model contains a ‘special regressor’ that is considered exogenous in addition to the endogenous variable and the instrumental variable (s). Also, the special regressor must be continuous and appears additively in the model. Since the exogeneity of the special regressor is assumed, the study used the household size for that role as it satisfied the other criteria.

The SRM was used because it performs better in the case of a binary choice model with an endogenous regressor compared to models such as the Linear Probability Model, Control Function Methods, and Maximum Likelihood Methods (MLM). For instance, unlike the MLM, the SRM allows for heteroskedasticity of unknown form in the model's error process (Baum 2012). This improves the robustness of the estimated results. The SRM employs an instrumental variable (IV) approach in its estimation which requires the use of an instrument that satisfies the correlation and validity constraints of the IV approach (Greene 2008). While we admit the difficulty in finding a valid instrument given the available data, the poverty status of the household was used as an instrument. The poverty status of the household strongly correlates with their receipt of remittance which further affects their choice of fuel. Since migration involves significant financial commitments, poor households are less able to sponsor their members to migrate compared to non-poor households. It follows that non-poor households are more often than not likely to receive remittances compared to poor households. The receipt of remittance has the potential to influence the choice of fuel used by the household because it improves its income.

4. Results and discussions

4.1 Distribution of households based on their choice of cooking fuel

This section sheds light on the choice of cooking fuel used by households in Ghana. The original variable comprising of the different types of cooking fuels was recoded into dirty and clean fuels. It tells the proportion of households choosing a particular kind of fuel for cooking in the past year. The distribution is presented in Table 3.

Table 3. Choice of cooking fuel by Ghanaian households.

Clean fuel	Frequency	Percent
No	50,501	85.82
Yes	8,345	14.18
Total	58,846	100.00

Source: Ghana Living Standard Survey (GLSS, 2017).

Table 3 shows that 85.82% of Ghanaian households do not use clean-cooking fuels such as LPG and electricity. This implies the majority of households rely on dirty-cooking fuels such as firewood, kerosene, and charcoal for cooking. Only about 14% use clean fuels for their cooking which is worrying and underscores the need for more households to shift to clean fuels. The high percentage of households using dirty fuels is appropriate for this study as we analyze how the receipt of income remittance can aid them to shift to clean fuels.

4.2 Influence of income remittance on the choice of cooking fuel by Ghanaian households

The purpose of the current research is to examine how the receipt of income remittance by Ghanaian households influences their choice of cooking fuel. This is against the background that the transition of household from the use of dirty fuels to clean ones reduce degradation and safeguard the environment. The results from the empirical estimation are presented in Table 4.

Table 4. Effect of income remittance on the choice of cooking fuel.

Clean fuel	Coefficients	Odds ratio	AME
Income remittance (ln)	0.267*** (0.029)	1.306*** (0.038)	0.00957
Household wage (ln)	0.989** (0.051)	2.689** (0.138)	n/a
Household size	−0.219** (0.014)	0.803** (1.011)	0.05
Location (rural)	−0.859*** (0.085)	0.423*** (0.036)	n/a
Poverty status (poor)	−2.804** (0.356)	0.061** (0.022)	n/a
_constant	−9.972* (0.500)	0.00467* (0.0023)	−0.378
Likelihood statistics
No. of observations	4,769	4,769	17,310
Wald chi2	737.75***	737.75***	262.49***
Pseudo R2	0.264	0.264	n/a

Notes: STATA output using data from the Ghana Living Standards Survey (GLSS, 2017), robust standard errors are in parentheses; ***, **, and * represent 1%, 5%, and 10% significance levels, respectively.

Results presented in Table 4 indicate that a percentage increase in the receipt of income remittance increases the likelihood for households in Ghana to adopt clean fuels. The propensity to use clean fuels increases by 0.00267 and by a factor of 1.31 for every increase in income remittance. The result for average marginal effect (AME) from the average index function also shows the receipt of income remittance increases the likelihood for households to use clean fuels by 0.0096.¹ This result is in line with a-priori expectation based on the fact that income remittance improves household income which pushes them up the fuel ladder where they consume clean fuels like electricity and LPG. In essence, this finding does not find support for the remittance-led pollution hypothesis in Ghana. It rather shows that the receipt of inward remittance can help reduce environmental degradation by aiding households to use clean fuels. According to Alem et al. (2016), the negative effect on environmental degradation due to the shift to clean fuels by households occurs via two channels. First, it reduces deforestation as they completely/partially stop the harvest of firewood for cooking. Second, since the use of dirty fuels involves some form of burning, the shift to clean fuels reduces both indoor and outdoor air pollution. This means as households’ income improves due to the receipt of income remittance increases their use of clean fuels which eventually reduces environmental degradation.

Additional results from Table 4 show a positive nexus between household wage and their propensity to adopt clean fuels. A percentage increase in the household wage increases the propensity to use clean fuels by 0.0098. That is, the odds to use clean fuels increase by 2.689 for every increase in household wage. Generally, higher wages increase the ability of households to afford clean fuels as they tend to be more expensive than fossil-based fuels. This finding also supports the energy ladder hypothesis that wage is a key factor that informs the decision of households to adopt a particular type of fuel. It, therefore, follows that higher wages make it more probable for households to shift to clean and efficient cooking fuels (Mensah and Adu 2015; Bofah, Appiah-Konadu, and Ngwu 2022).

Also, an increase in the size of the household reduces the likelihood for households to use clean cooking fuels. Households are less likely to adopt clean fuels by a factor of 0.8 for every increase in their size. Also, the probability to use clean fuel falls by 0.219 for every increase in household size. This implies larger households tend to use dirty fuels because of the high dependency burden on the few breadwinners. Holding all other factors constant, the household income is spread over a large number of people leaving little or no income which reduces their ability to afford clean fuels (Bofah, Appiah-Konadu, and Ngwu 2022). This result confirms the findings by Meng et al. (2021) who found households in Ghana to be more likely to use dirty fuels as the number of children and/or dependents increases. Per their findings, some larger households may not use any clean fuel at all.

Further results indicate that households in rural areas are less likely to use clean cooking fuels compared to those in urban areas. Precisely, the odds for households to use clean cooking fuels fall by 0.859 if they are in rural areas. This suggests that the use of dirty cooking fuels is prevalent among rural households mainly due to the low opportunity cost of using fossil-based fuels, low education, and low incomes (Nnaji, Ukwueze, and Chukwu 2012). For urban households, higher living standards make them more likely to use clean fuels. This is further supported by the high time and opportunity cost of using dirty fuels in urban areas which involves collecting and assembling charcoal or firewood to use (Hou et al. 2017).

Lastly, findings reveal that poor households are less likely to adopt clean cooking fuels. Specifically, poor households are 0.061 less likely to use clean fuels compared to non-poor households. As expected, poor households have low income, low education, and limited economic opportunities so it is difficult for them to afford clean fuels like LPG and electricity for cooking. Unlike non-poor households, poor households find fossil fuels like firewood relatively cheaper making them less likely to opt for clean cooking fuels (Martey et al. 2021).

In light of the findings, there is a strong connection between the living standards of households and their choice of cooking fuels and by extension their ability to transition and reduce environmental degradation. Households that have received income remittance, have a wage income, are non-poor, and reside in urban areas are more likely to adopt clean cooking fuels. On the contrary, large, poor, and rural households are less likely to choose clean fuels. It can therefore be said that an improvement in the living standards of Ghanaian households is important for them to use clean fuel and reduce environmental degradation. According to Statista (2023b), about 3 million Ghanaians live in extreme poverty with the majority of these people residing in rural areas. Besides, households who are better off can afford to send their people abroad and receive remittances to further improve their lives. Therefore, efforts to reduce poverty and improve livelihoods, especially in rural Ghana will be a step in the right direction to increase the use of clean fuels and reduce environmental degradation.

5. Conclusion, policy implications, and limitations of the study

5.1 Conclusion

The United Nations Sustainable Development Goals (SDGs) include the objective to ensure the consumption of clean and affordable energy by 2030. Consequently, national and international partners have initiated numerous efforts to discourage the use of fossil fuels by households in favour of clean fuels but this transition comes at a cost that can be budget-breaking. The literature on the determinants of household cooking fuels has ignored the potential of inward remittance to improve household income and assist them to shift to the use of clean fuels. We fill this knowledge gap by investigating the influence of income remittance on the choice of household cooking fuels in Ghana. The current research relies on household-level data from the Ghana Living Standard Survey (GLSS, 2017) conducted by the Ghana Statistical Service (GSS) using a logit model.

Based on the empirical findings, we conclude that the receipt of income remittance positively impacts the propensity of households to adopt clean cooking fuels. By extension, the receipt of income remittance helps in the fight against environmental degradation as household transition to clean cooking fuels and reduce their reliance on fossil-based fuels. We, therefore, find no support for the remittance-led pollution hypothesis in Ghana. Also, the research concludes that increases in the household wage increase the likelihood of households choosing clean cooking fuels but increases in the size of the household reduces this propensity. Moreover, we can conclude that poor and rural households are less likely to adopt clean-cooking fuels in Ghana, compared to non-poor and urban households. Consequently, the current research provides micro-level evidence in support of the negative effect of inward remittance on environmental pollution by influencing the choice of household cooking fuels.

5.2 Policy implications

This research has some implications for policymakers as far as energy transition in Ghana is concerned. First of all, poverty was found to be a barrier to the adoption of clean cooking fuels as the poor are more likely to use dirty cooking fuels. Government policies and strategies such as the Livelihood Empowerment Against Poverty (LEAP) must be revamped to lift Ghanaians out of poverty and encourage the use of clean cooking fuels.

Also, an increase in household wage improves the likelihood for households to adopt clean cooking fuels. The implication is that the more individuals earn a wage, the higher their chances of using clean cooking fuels as the opportunity cost of using dirty fuels increase. Policies to reduce unemployment will allow more individuals to earn wages which increases their propensity to adopt clean fuels.

Moreover, extra sources of income like inward remittance encourage the adoption of clean cooking fuels among Ghanaian households. This suggests that alternative sources of income can boost household income, and improve their living conditions which encourage the use of clean fuels and reduce environmental degradation. Households are advised to create alternative sources of income to improve their living standards which is vital for them to adopt clean fuels.

Lastly, the research entreats government agencies on the need to intensify education and awareness of family planning. This will help reduce their household size which is important in reducing their dependency burden and improving the quality of life necessary for the adoption of clean fuels.

5.3 Limitations of the study

This study concentrated on the choice of households between using clean and dirty cooking fuels as the dependent variable. It restricted the analysis to that of logit regression. We encourage future studies to explore the effect of income remittance on the specific fuel types with different methods to expand the frontiers of knowledge.

Another area for future studies is that the current research limited the discussion to just the effect of income remittance. However, there are different types of remittance and it will be interesting to know how the receipt of other forms of remittance like tradable goods influence the choice of households. The current data set has no information on this.

Although the GLSS has information on over 59,000 households, the cleaning and treatment of data to the relevant variables reduced the sample to just 4,765 households. To increase representativeness and advance knowledge, we suggest that future studies conduct specific surveys that include a lot more respondents for each of the variables of interest. This made it difficult to use an instrumental variable that satisfies the validity restriction of an IV method.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The AME was from the Average Index Function which is based on the special regression results which is presented in Appendix.

Appendix

Results from the special regression method (SRM)

. sspecialreg fuel_choice hhsize,kdens endog(lremittance_income) iv(pstatus)

Kurtosis of special regressor hhsize = 3.6195

42,554 observations trimmed: max abs value of transformed variable = 5.02 sigma

Instrumental variables regression Number of obs = 17,310

Wald chi2(1) = 262.49

Prob > chi2 = 0.0000

Root MSE = 17.328

————————————————————————————

fuel_choice | Coefficient Std. err. z P>|z| [95% conf. interval]

——————-+—————————————————————-

lremittance_income | 5.253801 .3242783 16.20 0.000 4.618227 5.889375

_cons | −39.52121 1.988533 −19.87 0.000 −43.41866 −35.62375

————————————————————————————

Instrumented: lremittance_income

Instruments: pstatus

Average marginal effects from average index function

fuel_choice

hhsize .00957096

lremittanc∼e .05028395

_cons −.37825607